How frequently are the osm-intl tiles rerendered to reflect changes to OpenStreetMap? Map improvements 2018#April 18, 2018, Special Update on Map Internationalization says it currently takes up to 48 hours, which I take to mean the time it takes for the extract or planet diff to be pulled in, as opposed to the tile rendering schedule. A change I made 15 days ago, correcting the name of a major city in Vietnam, appears at z10 and above but not at z9 or below. Is there a way to force-update a particular tile, similar to appending /dirty to URLs on the OpenStreetMap tile server?
Topic on Talk:Map improvements 2018
Sorry for the late answer (vacation). The process is as follow:
1) upgrade our local copy of OSM (postgresql database) every day (we are planning to increase this frequency)
2) regenerate vector tiles for any tile that has changed, triggered by the above update, but can take a few hours to complete 3) raster tiles are then generated on request, from the vector tiles and cached for up to 24h
There is no public ways to trigger steps 1) and 2). You can always bypass the cache by adding a dummy parameter to the URL to test if the issue is in 3).
All that being said, an update not showing after 15 days definitely sounds like a bug on our side! Could you please open a ticket on phabricator and tag it with "Maps"? Thanks!
We follow the standard practice of only regenerating high-zoom tiles based on updates. I've created task T194787 about regenerating low-zoom tiles.
I also ran a manual run to regenerate z0-z9 in both data centers, which should result in those tiles being fixed.
Even with a dummy parameter, I'm still not seeing the update as a result of the manual run.
Interesting question @Mxn. I'm pinging @Sbisson (WMF), who may be able to help you with an answer.
It's been over a month since the tiles were manually regenerated, but the edit still doesn't show up at z9. Would you mind taking another look at this, Pnorman? Thanks!
This is mostly fixed now, but this tile stubbornly refuses to update.
As my contract with WMF has finished, I'm not responsible for this anymore, and without the new style being deployed, the updated scripts aren't being redone. I suggest you open a phab ticket so the issue doesn't get dropped.
Hello @Mxn, we have worked on this issue for a while, as you can see here:
- https://phabricator.wikimedia.org/T194787
We are also working on some related tasks considering caching improvements, which I believe is the issue now:
- https://phabricator.wikimedia.org/T159977
I would reinforce what Pnorman said and ask you to create a phab ticket with the issue and detailed descriptions so we can track it.
Thank you!
Thanks, I opened phab:T202201 to track this specific case.